This blog is opinionated, unapologetically biased and all about cloud technologies. I will talk about cloud services such as Lambda, Step Functions, EventBridge, DynamoDB, QLDB, ECS and Fargate. I will talk about infrastructure such as Graviton, Nitro, Firecracker, Inferentia and Trainium. I may at times mention a technology that is Cloud Native under the hood yet compatible with traditional technologies; think of Aurora, Keyspaces or DocumentDB.
This blog is my personal opinion and does not represent my employer.
Why focus on Real Cloud Native services?
Real Cloud Native services are fully managed, rapidly innovating, come with high SLAs, are secure by default, have low RTO/RPO (for databases) and scale well beyond most customer's needs. These services are well-integrated and keep optimizing based on customer demand. And importantly, these services are well-optimized for low cost. They are pay-for-value and not bound to license fees.
How about lock-in?
This is worth a separate blog post that I haven't yet written. Here's a short version:
Lock-in is a multi-dimensional spectrum rather than a simple yes/no question. Your position on this spectrum depends on context, risk appetite and cost tradeoffs. Watch out not to be dogmatic on a single dimension only. Ask yourself: what is the total additional cost and complexity of a cloud agnostic technology across 5 years? What is the likelihood of actually switching to another cloud during these years? In comparison, what is the additional cost of switching if you accept some level of lock-in?
In my experience, the likelihood of switching tends to be lower than estimated. And any actual switches in year 5 look entirely different from what you imagined in year 0 due to technology advancements. I suggest emphasizing the best possible technology for your customers at the lowest cost (value for money) and considering lock-in as an important though not primary consideration.
For the purposes of this blog, that means Real Cloud Native technologies!